Spillways – ICOLD Bulletin on Cost Savings
3.3. Spillways
The chapter refers to high and to low dams.
For several reasons the design of spillways represents a significant opportunity for cost saving. Spillways not only represent a significant direct costs, by virtue of their design and possibly by their impact on the overall project layout, but also on the potential cost of lost storage in the case of ungated spillways. Over the last 50 years the accepted tolerance of dam failure risk has decreased while at the same time the estimated value of extreme floods has increased. Many future dams will be built in Asia and hence close to warm seas, giving the potential for very large floods and run-offs per km² of catchment area. Climate warming may also increase floods in the future. The discharges required for future spillways are therefore likely to be higher than in the past, for similar catchment areas. ICOLD has devoted many studies to this problem, and especially:
– Question 79 of the 2000 Congress
– Part of the Question 84 of the 2006 Congress, and especially the General Report (pages 1566 to 1579).
– ICOLD Bulletins 82, 108, 109, 125.
Cost saving opportunities may be based on a number of new technical solutions developed worldwide and presented below. Preliminary comments on spillway requirements, design criteria, past failures and floods evaluation seem useful for better evaluating these solutions.
3.3.1. Spillways requirements
Many spillway designs, in the past and even now, have been based upon standardized methods. These can involve the evaluation of a “design flood” of a given annual probability. These are often chosen according to regulations or tradition, such as a 1 in 1 000 years event, with a requirement to discharge this flood without any damage, using a fully gated or fully ungated spillway. During such floods the reservoir level is often kept below the dam crest by a significant freeboard. This approach may be reviewed considering at least three other factors.
– The risk of failure by more exceptional floods.
– The risk of gates jamming and especially of all gates jamming. This has been the reason for over 20% of flood induced failures at gated dams.
– The risk from failure of upstream reservoirs, including natural reservoirs, such as glacial outburst floods in regions like the Himalayas.
In addition, spillway facilities designed essentially for the dam safety could often be also used for improved siltation management and the reduction and mitigation of floods downstream.
3.3.2. Design criteria and methods
Design criteria based upon “design floods” and significant freeboard have two main drawbacks:
– The true failure probability is unknown. It may be higher for a large gated dam or a morning glory spillway than for a small ungated dam. The true margin of safety is based upon the freeboard i.e. upon local wind speed because the freeboard is designed according to the relevant wave heights. The uncertainty on the true risk is increased by the significant uncertainty in evaluating the value of floods with a 1 in 1 000 annual probability.
– This design approach can reduce or prevent the utilization of alternative low cost solutions for increasing the spillage capacity at extreme floods. This drawback is often overlooked because such arbitrary criteria focus only on performance requirements and not on overall cost.
An approach advocated by ICOLD since 15 years, especially in bulletins 82, 108, 125, 130, suggest the use of a “Check Flood” or “Safety Check Flood” of much lower probability. This would require a reduced margin of safety and reflect a higher reservoir level. Indeed it is likely to be a reservoir level close to that which could cause dam failure. Limited damage would be accepted for this “check flood”. The alternative names of “check flood” and “design flood” may be confusing. The most important flood dominating the final arrangement may well be the “check flood” and not the so called “design flood”.
3.3.3. Dams failures by floods
For evaluating the reservoir level acceptable for the “check flood”, it is useful to review the conditions of past failures. They are very different for embankment dams, concrete dams and masonry dams:
– Over 2% of all embankment dams built before 1930 and of embankment dams built in Asia between 1950 and 1980 failed by floods (ref. ICOLD Bulletin 109, p.27 and appendix 1) i.e. an annual failure rate over 1 in 1 000. All these dams had been built without heavy construction equipment and often with poorly compacted earthfill. The annual failure rates by floods since 1980 are lower, in the range of 1 in 10 000 for very large dams but still close to 1 in 1 000 for smaller ones. The relevant fatalities, which have been high for some past failures, are now reduced by weather forecasting and improved telecommunications. The flood failures of embankments are usually due to crest overtopping by more than 1 m for rockfill dams, by about 0.5 m for well compacted clay and by much lower values where small size gravel is concerned. Typically erosion gulleys are created in the downstream toe or slope and reach progressively to the dam crest. The resulting breach then opens quickly with a local discharge (in m3/s) up to H2,5 where H is the local dam height (in m). This breach widens slowly for cohesive fill and quickly for rockfill or gravel. The acceptable reservoir level for the “check flood” may thus be close to the crest level for a rockfill dam or a small earthfill dam, say, 0.5 or 1 m below crest level for a cohesive dam crest designed to withstand some overtopping by waves, and at least 1 m below crest for gravel fill.
– The discharge from failure may be much higher than the natural flood but if an embankment dam fails for a very exceptional flood, the natural discharge may not be much increased especially if the dam is not high where the breach occurs and also if the dam material is cohesive.
– No failures by flood have been reported for the 2 000 concrete gravity dams higher than 20 m but several have been reported for such dams between 10 to 20 m high, usually by sliding. This may be explained by the much higher relative impact on a low gravity dam of stability from a reservoir level increase by few metres and also due to uplift increases associated with a downstream level increase. Several high concrete dams have safely withstood significant overtopping. Two small concrete arch dams failed by overtopping and the scour of the poor quality rock abutments.
– The rate of failure by floods of masonry gravity dams has been over 1%, including dams 40 m high. Some failed before overtopping, several failures were in the masonry itself, possibly due to low masonry tensile strength.
Failures of concrete or masonry dams are instantaneous and cannot be foreseen very precisely. The breach is wide, often in the range of five times the dam height and the failure of a concrete dam is thus less likely but more dangerous than the failure of an embankment dam for a same reservoir volume. The risk from existing masonry dams may be high.
3.3.4 Flood evaluation
The methods for flood evaluation may be different for the “design flood” and for the “check flood”. Probabilistic methods based upon flood and rainfall data are used for evaluating the “design flood”. The uncertainty in such evaluations increases with the return period being evaluated.
The “check flood” should be chosen with a very low probability, such as 1 in 100 000 or perhaps the Probable Maximum Flood (PMF). Probabilistic methods of evaluation may also be used but their reliability for such exceptional floods is questionable, especially because extreme floods may be caused by different climatic events than those which cause frequent flooding. The deterministic method of the PMF seems more reliable but results may vary significantly with the experts involved in evaluating it. It is especially difficult to evaluate the PMF of very large catchment areas which contain variable climatic conditions.
Approximate evaluations of extreme floods may be easier for catchment areas under few thousands km² subject to heavy rains, i.e. for most future dams. In these cases the extreme flood is usually caused by a same rain event in the range of 0.5 m depth in a short time over the whole catchment area. The extreme flood duration varies between 3 and 12 hours according to the catchment area. The soil conditions have little impact on the maximum discharge which is essentially linked with the area of the catchment and the regional climate. A useful reference (presented in ICOLD Bulletin 125, p. ) is the curve of extreme discharges reported worldwide according to the catchment area. They are broadly:
These may be roughly represented by two simplified formulas (in m3/s)
For S < 300 km² : q = (S/300)0,8 x 10 000
For S > 300 km² : q = (S/300)0,4 x 10 000
The specific extreme discharge qs may thus have been often underestimated in the past even for catchment areas of 100 or 1 000 km².
Natural specific flows were much lower for most dam failures than the envelope curve above but reached 90% of this world reference for Banquiao and Shimantan (China 1975), 70% for Machu (India 1979) and 50% for Molare (Italy 1935). Such accidents caused many fatalities. A simplified additional evaluation of the check flood may thus be obtained based upon the catchment area, and the maximum reported regional rain compared to the world maximum rain. The extreme world rains have reached 1.50 m in 24 hours and 1 m in 12 hours. Some limited adjustment may be added to allow for the shape and slope of the catchment area.
Such additional evaluation, as with other methods, is questionable, but is based upon well established and known data. It is inexpensive to evaluate and is at least reliable for comparing the safety of dams in a same climatic area. The volume of such extreme worldwide floods for rather small catchments is a high percentage of the possible rain in a few hours. The volume per km² is similar for such catchments, and may be in the range of 0.50 m over the catchment.
For small catchment areas, the ratio between a flood such as the P.M.F. and a flood of probability 10-2 is usually high and may be well over 3 or 4. This ratio is usually lower for very large catchments because extreme events tend to apply only to a part of such catchments. Uncertainty will increase with climatic changes.
3.3.5. Low cost solutions for the “check flood”
The studies below consider cost savings based upon the approaches suggested above.
The “check flood” discharge may be divided into two parts: q1 which is the discharge of the “design flood” and q2 which is the difference in discharge between the “design flood” and the “check flood”. The sum of q1 + q2 should be as high as possible.
The direct cost (c1) per m3/s for discharging q1 is usually high for gated spillways and indirectly high, for free flow spillways due to the associated loss of reservoir storage corresponding to the nappe depth of the “design flood”. Such loss is often in the range of 10.000 m3 per m3/s and this loss of storage can be a significant part of the live storage. The value of c1 is thus considerable, possibly over USD 10.000. The cost c2 per m3/s for discharging q2 may be much lower because q2 may use a higher reservoir level and cause some acceptable damages. The total cost, (c1 q1 + c2 q2 ) should be as low as possible. If it is difficult to reduce c1, there are two ways of cost savings: essentially to reduce c2 or to somehow reduce q1 to some extent and to increase q2 accordingly.
Many solutions may be used for discharging the “check flood” at low cost and especially the flow q2 beyond the design flood. Some have been overlooked in the past where the “design flood” was assumed to be the main criterion. The solutions available may be different for embankments or for concrete dams or may apply to both.
Most earthfill dams have slopes in the range of 1 on 2 or 1 on 2.5, or even flatter. In many cases the slope is kept the same throughout, including the upper part where freeboard is chosen according to possible wave action. It is often possible to keep most of the dam body unchanged but to steepen just the upper sections to, say, 1 on 1.5, in order to raise the crest by 1 or 2 m. Such a raising of the crest by 1 m requires about 20 m3 of earthfill per metre of dam length at a cost of a few hundreds USD. Raising the nappe depth of the spillway by 1 m typically increases the discharge per metre of spillway by 5 to 10 m3/s. If the dam length is 5 to 10 times the length of the spillway, the cost for increasing the discharge by 1 m3/s is well under USD 1 000, often under USD 500, even allowing for some extra costs in the downstream waterway and for minimum waterproofing and waves protection of the crest. It is also possible to add a parapet 1 m high in concrete or gabions.
It is usually possible to raise Concrete Faced Rockfill Dams by up to 5 m by means of a concrete or RCC structure at moderate cost in terms of both the structure required and the additional spillway discharge produced by the raising.
Long embankment dams usually include a significant length where the dam height is in the range of 5 or 10 m. It may prove economically advantageous to raise just the higher sections of such a dam by about 1 or 2 m by local crest steepening, while leaving the lower height sections of dam at the original level. Any breaches which occur at these lower sections will have less consequence than for the higher sections of dam. Indeed it may be possible to specifically design such low sections as earthfill fuse plugs incorporating simple associated structures to limit their development in the event of failure.
Various alternatives have been used for spilling over embankment dams. Two solutions may have a promising future, at least as auxiliary or emergency spillways. Beyond a gated spillway to discharge the design flood, it is cost effective for dams between 5 and 10 m high, to set the crest level at 0. 5 m above the normal reservoir level and to line the downstream slope with RCC placed in layers about 3 m wide. For a 3 m nappe depth, corresponding to most of the main dam freeboard, this requires about 5 m3 of RCC. per extra m3/s, at a cost in the range of USD 500 per m3/s.
Ungated reinforced concrete spillways have been placed upon some Concrete Faced Rockfill Dams with specific discharge per metre length of spillway of 20 m3/s. In some cases these represent the only spillway provision. Alternatively it may be preferable to use such spillways as emergency spillways in addition to a traditional gated spillway set on the valley flanks. It may also be possible to increase the specific discharge of such spillways and enhance available reservoir storage using inflatable gates, fuse devices or labyrinth weirs.
High embankment dams in narrow valleys may require costly lined and gated tunnel spillways with expensive downstream structures. Such expensive spillways could be limited to the discharge of the “design flood”. It may be possible to accommodate more exceptional floods using unlined, high-level tunnels operating at much lower velocities. The upstream entry control could feature fuse devices and downstream energy dissipation provision could be minimised provided the extent of any associated scour was understood and acceptable. The relevant cost per m3/s may be one third of the cost for the main spillway. Such tunnels are in fact similar to temporary diversion tunnels.
Many concrete dams may withstand overtopping by exceptional floods, even with significant specific flows. This can be accepted provide that downstream scouring is avoided or acceptable and the structural stability is confirmed based on the higher reservoir level. The corresponding exceptional extra loads should however not be combined with maximum possible seismic loads. For reducing the maximum reservoir level during extreme floods, it may be preferable to reduce the freeboard and to choose a steel parapet instead of a concrete parapet. It is possible also to use fuse concrete plugs as presented below.
The optimal cross section of a gravity dam may not be the traditional one but rather one with an inclined upstream face (see ICOLD Bulletin 109, appendix 2). For significant overtopping a symmetrical section may be preferable to one with a conventional, vertical upstream face. For gravity dams, scouring by exceptional floods may be limited by using minimal and low cost protection at the downstream toe. Stepped downstream faces are especially efficient for the dissipation of overflows up to specific discharges per metre length of spillway of approximately 10 to 15 m3/s, i.e. for ordinary floods. It may be possible to accept some erosion damage downstream for more extreme floods provided, again, that it’s extent and implications are understood and acceptable.
Low cost solutions for discharging the “check flood” at many embankment or concrete dams may also include the utilization of various fuse devices. They may be used as auxiliary spillways, the design flood being discharged by a basic gated spillway, or they may be overtopped by the design flood. They open partly or totally for exceptional floods and it will take some days, weeks or months to replace them with some associated cost or loss but with an annual probability in the range of, say, 1 in 100 or 1 in 1 000.
Many fuse plug solutions have been used for small dams and also for large ones. For instance earthfill fuse plugs have been used at about 100 large dams, mainly in China and in the United States about 20 years ago, usually for discharges of some thousands m3/s. They fail by erosion of materials and breach widening. They require specific topographic conditions and there are questions about the relevant associated downstream hydrographs and also about their long term reliability. As they are designed for very exceptional floods, there are very few examples of them having successfully functioned.
Another solution developed over the last 15 years and used in about 10 countries is based upon gravity concrete or steel elements tilting in sequence for different values of reservoir level. In this fusegate solution, uplift is created under each element for a precise reservoir level. Fifty spillways have used this solution. Such fusegates may be designed for overtopping by the design flood before tilting. They may have a labyrinth layout in order to reduce the corresponding nappe depth before tilting and thus reduce the loss of reservoir storage. They have been used for both small and large spillways, up to 40 000 m3/s. Most have been overtopped; some elements have tilted for dozen of them as designed.
Two simpler concrete fuseplugs which open by tilting have also been studied and model tested. In the first, the plugs tilt just before or just after overtopping and are designed to avoid uplift. Their thickness is about half their height and they can be used to replace the upper parts of standard concrete gravity dams. This minimises their cost as in such a situation they require no additional concrete. In the second solution, the concrete fuseplugs are overtopped by the design flood before tilting. They are designed with full uplift. Their cost for a new free flow spillway would be low and they may be designed for the PMF. However, they do not have the same labyrinth shape as fusegates and the associated tilting level is less precise. They may also be used for upgrading existing free flow spillways (see appendix 3).
Many other solutions have been used, such as flashboard in the U.S. for small spillways. These are wooden boards standing against vertical steel pipes. Their cost is very low but they may be vulnerable to damage by floating debris or to willfull damages.
3.3.6. Low cost solutions for the design flood or the check flood
Cost savings may be obtained by improvement of free flow spillways or of gates, or associating both solutions. Other savings apply to waterways and downstream structures.
Most existing large dams have ungated spillways (ICOLD Bulletin 83, p.33) i.e. a great majority of spillways for design floods under 1 000 m3/s and a significant part of larger ones. Free flow spillways are usually for catchment areas under some hundreds km²; the time to peak of such floods is usually 3 to 6 hours. Free flow spillways are very safe for managing such floods but suffer low specific discharges, in m3/s per metre length of spillway of about 2,2 H1,5, where H is the nappe depth in metres.
The loss of storage corresponding to this nappe depth for the design flood is often 20 to 50% of the live storage of irrigation dams. For a reservoir area S (km²) a spillway length L and a nappe depth usually in the range of 2 m. the loss of storage (m3) is 2 x S x 106 and the flow 2,2 x L x H1,5 is close to 6 L (m3/s); the loss of storage per m3/s is thus :
For instance, for S = 5 km² and L = 100m (discharge close to 600 m3/s) the loss of storage per m3/s is 15 000 m3.
This loss may be reduced by increasing the length L of the spillway, for instance with side spillways along a bank or by duck bill spillways. It may also be reduced by increasing the discharge for a same nappe depth using labyrinth shapes instead of the traditional Creager cross section. About 100 spillways use this solution worldwide, generally with vertical reinforced concrete walls and with a trapezoidal labyrinth layout placed on a horizontal bottom, usually in a flat bank. The developed length of walls is often in the range of 4 times the spillway length, the nappe depth about 50% of the wall height. This solution has been mainly used for spillways of some hundreds m3/s with walls 3 or 4 m high; but sometimes for discharges up to 15 000 m3/s with walls up to 10 m high. Usually the discharge is about double that of a Creager Weir discharge for a same nappe depth. Such spillways are easy to build and have operated well, some of them for over 50 years.
However these traditional labyrinth designs have 3 drawbacks. Vertical walls are not the best hydraulic shape for accommodating horizontal flows and the performance is thus reduced, especially for large discharges. In addition the necessary quantity of reinforced concrete for increasing the discharge by 1 m3/s is close to 2 m3. But the most important drawback is their length in the flow direction which can be about 3 times the wall height. This prevents their utilization on gravity dam crests, i.e. on most dams or traditional spillways.
The performance and cost efficiency of such labyrinth spillways has been significantly increased by an analysis of their drawbacks and by hydraulically model testing alternative shapes. These have optimised their hydraulic performance and also with due regard to considerations of structural design and construction. The relevant options are presented in more detail in appendix 2 to this bulletin and summarized below.
New shapes of labyrinth weirs, using inclined shapes and hangovers (including so called Piano Keys Weirs) and a rectangular layout instead of a trapezoidal shape have been studied in 5 countries in recent years and appear very promising. They may multiply by 3 or 4 a Creager Weir discharge for the same overspill depth. They require 0.5 to 1 m3 of reinforced concrete per extra m3/s of discharge and may be placed upon most existing free flow spillways or future gravity cross sections. Their discharge may be easily checked by simple model tests on existing laboratory flumes. They may be even more beneficial in countries with low cost labour and hence low cost reinforced concrete (see appendix 2).
Associating two spillways
For the great majority of existing dams, the extreme flood is discharged through either a fully gated spillway, or a free flow spillway. Associating two spillways may often be the best solution both for dam safety and for a better management of floods and siltation.
For instance, the best solution for large discharges, say over 10 000 m3/s, may be a gated spillway for the design flood and an ungated spillway with fuse devices or labyrinth weirs for extra discharges up to the check flood. The gates will usually be large surface sector gates (possibly with some orifice gates, 20 to 50 m under the reservoir level for flushing sediments). The auxiliary ungated spillway will be much less expensive than the cost of increasing the gated spillway capacity and will provide an emergency safeguard in the event of all gates jamming. A great advantage of large gated spillways is achieving specific flows up to 150 m3/s per m of spillway. Emergency spillways may reach 100 m3/s per m of spillway with fusegates and 50 m3/s per m with labyrinth weirs using the freeboard depth.
Smaller discharges are usually made by free flow spillways which may be improved by the use of labyrinth weirs or fuse devices. It may also sometimes be useful to use also some gates for improving siltation management or flood mitigation. For instance irrigation dams with high siltation problems usually store a part only of their annual flow. Low gates may be used for keeping the reservoir empty during part of the flood season, thus sluicing most of sediment, with the storage being completed in the second part of the flood season.
For irrigation dams storing most of the annual flow, associating a labyrinth weir for the extreme flood and a bottom gate for spilling the annual flood also allows the downstream peaks of many intermediate return period floods to be mitigated.
In these two last cases there is no need for permanent operators and mismanagement or jamming of gates have limited impacts.
Cost savings in gates
There has been much progress in gate design over the years. Further progress may be related to improving their safety, operation and maintenance rather than in cost savings from improved structural designs. The most effective solution is usually radial gates. They may be very large, up to 400 m² per gate leaf and corresponding civil engineering will be a large part of spillways cost. They are also often preferred for low gates as well as being quite easy to use for heads of up to 50 m.
Higher heads require great care with problems such as cavitation and erosion by sediments: but the need for high discharges at heads greater than 50 m is unusual.
Flap gates are common for surface spillways with specific discharges per metre length of spillway of under 20 m3/s. Inflatable gates have also been used for such discharges but mainly for increasing the storage of existing dams. Flap gates supported by air bags (the Obermeyer System) have been developed since 1995 and have some advantages of both systems.
For avoiding the cost of permanent operators, gate automation has been developed in various countries, sometimes associated with computerized analysis of discharges and spillway management. Such solutions may be very cost efficient but incidents, especially under the conditions associated with exceptional floods, cannot be totally avoided. It is thus advisable, either to keep operators available during floods, or to use automation essentially for small discharge adjustments with limited impact of possible wrong operation. For instance, the upper part of large surface sector gates may be used as flap gate and use automation for 10 or 20% of the maximum discharges.
The risk of downstream fatalities by rapid gate opening and hence an associated rapid increase in discharge should always be born in mind when designing operating systems. The risk is especially high if the prevailing discharge is low or nil before gate opening.
Generally, reservoirs controlled by free ungated, spillways will tend to attenuate floods with some of the incoming flood rise going into storage rather than passing downstream. However it should also be noted that where such a reservoir is partially empty the early part of the incoming flood will fill the reservoir and outflows will only occur when incoming flood discharges have built up to high levels. Under such circumstances the rate of increase for downstream flows can be higher than would have occurred at the start of a natural river flood.
Waterways and downstream structures
Air injection or entrainment on high velocity waterways has been used for some 20 or 30 years as a relatively low cost means of avoiding damage due to cavitation. Reduction of downstream erosion may be also obtained by:
– Mixing water and air in labyrinth spillways.
– Using stepped downstream faces on normal sloping spillways. The impact is reduced if the nappe depth of the free flow spillway is higher than the step height and some improvement could probably be made if discharges are broken-in a three-dimensional manner, perhaps using baffles, rather than just 2 dimensionally.
– Associating labyrinth weirs with stopped spillways.
3.3.7. Choosing “design floods” and “check floods” based upon simplified cost analyses
Three key elements are often overlooked in regulations or risk analysis: the uncertainly of flood evaluation, the risk of incorrect gate operation and the costs. Two examples are presented below for illustrating their importance on the choice of designs and in the selection of return periods for both the design flood and the check flood.
– The first example refers to a gated earthfill dam in a catchment area of 1 000 km², with a storage of 100 hm3 and a PMF of 7 000 m3/s, half of the world maximum for such an area. The flood evaluation for various probabilities may be (as an example) in the range of:
The calculations below may be easily adjusted to other flood evaluations.
A traditional design may assume a gated spillway for a design flood of 3 000 m3/s and a probability of 1 in 1 000. The relevant cost of the spillway being at least USD 5 000 per m3/s, giving a total of USD 15 million.
During an assumed dam life of 100 years there is a probability of about 10% of a higher discharge but the probability of failure is reduced by the freeboard, for instance to 2%. As there is, and there will always be, a serious uncertainty in flood evaluation, the true probability may well be between 0.5% and 5%. There is an additional risk of failure, linked with gates jamming, in the range of 10-4 per year or 1% in 100 years. Over 0.2% of existing gated embankment dams have failed for this reason.
The true risk of failure for such a dam in a century may thus be between 1 and 5%. Regulations imposing a design flood of 10-5 instead of 10-3 will increase the cost by 2 000 m3/s x USD 5 000 = USD 10 million, but the true risk of gates jamming is not much reduced. Furthermore it is doubtful if human risk is globally reduced if the added risk of construction worker fatalities for USD10 million of extra work were to be compared with the reduction of the risk from dam failure. Embankment dam failure risk by floods and operator error is now less than before due to better weather forecasting and telecommunications.
The cost per m3/s of emergency spillways for the check flood may be (as per figures in 3.1.5. above) between 10 and 50% of the cost per m3/s of the basic spillway. An alternative may thus be to choose a gated spillway for a design flood of 2 500 m3/s and a return period of 1 in 500 years, and an emergency free flow spillway, for instance with labyrinths or fuse devices or spilling over a low part of the embankment lined in RCC, for a further 2 500 m3/s, giving a total check flood capacity of 5 000 m3/s with a return period of 10-5.
Assuming a cost per m3/s of emergency spillway to be 20% of the cost of a gated spillway, the cost of the emergency spillway will be 2 500 x 20% x 5 000 = US$ 2.5 million, i.e. about the same as the saving from reducing the gated spillway capacity by 500 m3/s x 5 000 US$. The failure probability from gate jamming is reduced to effectively zero, thanks to the emergency spillway, and the failure probability in 100 years is 10-5 x 100 = 10-3, i.e. 1in 1 000. Even with the uncertainty in flood evaluation the risk is much lower than with the previous solution and the cost is reduced by USD 10 million.
– Another example applies to an ungated earthfill dam in a catchment area of 100 km² with a PMF of 2 000 m3/s (half of the world maximum for such an area) and flood evaluations according to return period in the range of:
If the design flood is chosen as 800 m3/s, a return period of about 1 in 500 years, and a spillway with a traditional Creager profile weir, a spillway length of 100 m will be required for a nappe depth of 2.5 m. If a freeboard of 1.5 m is added the failure probability becomes quite low. Overtopping such a crest by 0.5 m would require a nappe depth of 2.5 + 1.5 + 0.5 = 4.5 m, in turn requiring a discharge close to the PMF. Regulations imposing a design flood of 10-4 would probably be expensive and unhelpful.
An alternative may be to choose the PMF as the check flood, a 1 in 100 year flood as the design flood and to use a 50 m long labyrinth spillway in place of the 100 m long Creager profile. The extra cost for a labyrinth weir, say 500 m3 of reinforced concrete, will be much less than the cost saving in spillway length. The design flood of 600 m3/s, or 12 m3/s per metre length of labyrinth weir will require a nappe depth of 1.5 m, i.e. 1 m less than the previous case and hence also 1 m more storage. With a same freeboard of 1.5, and the same discharge criteria for failure, a total nappe depth of 1.5 + 1 + 0.5 = 3.5 m will be required giving about 2 000 m3/s, the same as for the basic solution.
The safety may be further improved, or the cost reduced by steepening the dam crest. It seems thus possible, modifying the design criteria and using new solutions, to increase significantly the storage and the safety while at the same time reducing costs.
3.3.9. Non structural measures
ICOLD Bulletin E 02 published in 2000 has analysed non structural measures which may be often more cost effective than extra structural investments, especially for flood management. They include:
– For gates: maintenance, redundancy of operating devices and training of operators,
– For flood mitigation: management scenarios and alarm systems.
Many studies, theories and regulations refer to the possible human fatalities from dam failures by floods and many investments have been made accordingly. This problem has been and remains very serious. Some comments may however be useful to consider:
– Most fatalities (possibly 90%) from past failures of embankment dams by floods would have been avoided by modern forecasting of weather and floods and by the use of modern telecommunications.
– Worldwide the future risk of fatalities from dams and floods may globally be higher due to incorrect gate operation for low discharges or from releasing floods with probabilities of 1 in 1 000 or 1 in 10 000 so as not to endanger the dam, than from a very reduced probability of dam failures. It may be sometimes more cost effective to devote monies to flood mitigation, to alarm systems, to operator training or to gates maintenance than to spillway upgrading. The risk of workers fatalities during extra construction works should also not be overlooked when assessing over risks of fatalities. The upgrading works may also sometimes cause temporary extra risks for the dams.
3.3.10. Conclusion for spillways
Traditional spillway design criteria may warrant review. There are many possibilities for low cost improvements in safety or efficiency, based upon lessons from past failures and analyses of prevailing future conditions.
The impact of costs and uncertainties in flood evaluation should be better taken in account. It may be more cost effective and also safe to choose a very rare “check flood” and to accept a “design flood” with an annual probability of 1 in 100 to 1 in 500. There are many options for reducing cost but there is no standard solution given the extreme diversity of dam types. Savings in investment thus require more to be spent on design for analysing various solutions beyond traditional ones. For dams upgrading, the design cost for analysing many solutions may be a significant part of the cost of upgrading works.